Neuronal specification of direction and distance during reaching movements in the superior precentral premotor area and primary motor cortex of monkeys

1. Single-unit neuronal activity was recorded in the primary motor and superior precentral premotor areas of two rhesus monkeys during an arm reaching task. The task involved moving a cursor displayed on a video terminal using a draftsman's arm-type manipulandum. From a centrally located start box the animal was required to move to 1 of 48 target boxes at eight different directions (0-360° in 45° intervals) and six distances (1.4-5.4 cm in 0.8-cm increments). Both direction and distance for the upcoming movement were unpredictable. 2. The activity of 197 arm movement-related cells was recorded and evaluated for each of the 48 targets. Histological examination showed the cells to be primarily in the primary motor cortex or in the premotor area around the superior precentral sulcus. Each cell's discharge was aligned on movement onset and averaged over five trials for each target. Movement kinematics including hand path velocity were also determined. The task time was divided into three epochs, a premovement period (PT), a movement period (MT), and total time (TT = PT + MT). For each epoch the average firing was correlated with the direction and distance of the movement using various regression procedures. 3. An analysis of variance (ANOVA) showed that the majority of neurons were modulated significantly by movement direction in each of the three time periods, PT (73.7%), MT (68.3%), and TT (78.5%). The relationship of the firing to direction was fit to a cosine tuning function for each significantly modulated cell. In 86.3% of the cells the firing was correlated significantly with a cosine function of movement direction in TT. A cell's preferred direction varied little for different movement distances. The mean difference in preferred direction for the smallest possible change in distance (0.8 cm) was 12.8 ± 11.4° (SD) and 17.1 ± 14.7° for the largest change in distance (4.0 cm). 4. Correlation analysis revealed that the activity of the majority of cells was modulated significantly by distance along at least one direction in each of the three time periods, PT (46.8%), MT (68.8%), and TT (67.7%). Subsequently, a univariate linear regression model was used to quantify a cell's discharge as a function of distance. For the regressions of firing with distance with a statistically significant correlation (r > 0.8), the mean slope was 3.59 ± 0.17 spikes · s^-1 · cm^-1 for the total time. The existence of a significant distance modulation was not invariably correlated with a cell's preferred movement direction. There was only a slight increase in the likelihood of a significant correlation with distance for movements close to a cell's preferred movement direction. Also, the average slope for distance was not greater for movements close to a cell's preferred direction. 5. A two-way ANOVA was used to test for firing modulation with respect to direction and distance and to examine whether interactions between these two parameters exist. Significant modulation with either direction, distance, or 'interaction' between these two parameters occurred in 130 of 153 cells evaluated. The interaction terms in the model are equivalent to target position. Next, a multivariate regression model was developed to relate the cell's discharge with direction and distance. A proportional reduction in error (PRE) approach was used to define the terms in the model. The six required parameters included distance, a sine and cosine function of direction, and target position. The model yielded a significant fit (P < 0.05) for 119 of the 130 analyzed cells with a mean R² of 0.61 ± 0.18. 6. Cells' responses were analyzed on the basis of the partial R² obtained for direction, distance, and target position. In 37 cells only the direction term contributed significantly to the fit and in four cells only distance was significant. In 78 cells either the partial R² for both distance and direction was significant or the partial R² for target position (interaction terms) was significant. In addition, a temporal component to the correlation was observed, the discharge during the premovement period was more likely to be significantly correlated with direction, whereas the discharge during the movement was correlated with both direction and distance.